A Thermodynamic Mirage: Finite Measurements Can Hide Infinite Irreversibility

Constructs a quantum system where a realistic photon counter reports finite entropy production while the true microscopic irreversibility is literally infinite. A clean no-go result for experimentalists.

Imagine you're standing at a river and you want to measure how fast the water flows. You drop a leaf and time it. The leaf gives you a number — a real, positive, finite number. But what if the river is actually an infinite waterfall just below the surface, and your leaf can only float on top? That's the core mechanism here: coarse-grained measurements — your leaf — can report a finite, meaningful signal of irreversibility while the actual microscopic entropy production diverges to infinity underneath. The committed claim: there exists a physically realizable quantum system where a standard photon-counting measurement yields a finite, strictly positive observed entropy production rate, even as the true microscopic entropy production is infinite. This is not a pathological toy model. The system is a degenerate parametric oscillator below threshold — a workhorse of quantum optics — coupled to a thermal bath through repeated bosonic ancilla interactions. The infinite irreversibility arises because, in the vacuum limit, forward and reverse trajectory measures have disjoint supports: absorption is forbidden, so certain reverse trajectories simply cannot happen, making their likelihood ratio blow up. The measurement apparatus is deliberately realistic: a photon counter with finite collection efficiency, gated time bins, and independent dark counts. These are not artificial imperfections — they are standard features of any real photon detector. The dark counts are the crucial ingredient. They act as false positives that paper over the support mismatch between forward and reverse processes. Because both trajectory laws now share a common support (every detector outcome has nonzero probability under both), the measured relative entropy stays finite. The detector literally cannot see the infinity. The mathematical result is tight. For fixed observation time T and detector parameters, the observed relative entropy is uniformly bounded as bath temperature T_b → 0, and remains strictly positive at vacuum. The hidden contribution scales as Σ_hid(T, T_b) = T · Q̇_0 / (k_B T_b) + O(1), where Q̇_0 > 0 is the stationary vacuum heat current. This is a 1/T_b divergence — it blows up as the bath cools — while the measured part stays put. The separation between what you see and what's there grows without bound. This matters because entropy production bounds are the backbone of stochastic thermodynamics. Experimentalists routinely use coarse-grained measurements to infer thermodynamic costs, efficiency bounds, and irreversibility in molecular motors, quantum heat engines, and biological systems. The standard inequality says measured entropy production ≤ true entropy production. This paper shows the gap can be not just large, but infinite — even when the measurement is informative (strictly positive, not zero). That's a sharper no-go than previous work, which focused on finite hidden contributions. The architecture is analytical, not computational. This is a mathematical construction and proof within the framework of quantum Markov semigroups and trajectory thermodynamics. The oscillator-bath model uses repeated-interaction dynamics with explicit Kraus operators, and the detector model is specified in full detail. There are no numerics to validate — the result is a theorem. The proof strategy is to bound the measured relative entropy above by exploiting the dark-count probability floor, then show the hidden part diverges by computing the vacuum heat current exactly. The paper is submitted to Physical Review E, which is the right venue for a rigorous stochastic thermodynamics result. It's a single-author paper, which in this subfield signals either strong mathematical confidence or insufficient peer pressure during drafting. The construction is clean enough that the main vulnerability is not error but scope: the result holds for this specific model class, and the question of how generic the phenomenon is — whether infinite hidden entropy production is typical or requires fine-tuned support mismatches — is left open.